Identifying the character of ferromagnetic Mn in epitaxial Fe/(Ga,Mn)As heterostructures
M. Sperl,1F. Maccherozzi,2F. Borgatti,3A. Verna,4G. Rossi,4,5M. Soda,1D. Schuh,1G. Bayreuther,1W. Wegscheider,1 J. C. Cezar,6F. Yakhou,6N. B. Brookes,6C. H. Back,1and G. Panaccione4,
*
1Institut für Experimentelle Physik, Universität Regensburg, D-93040 Regensburg, Germany
2Soleil Synchrotron, L’Orme des Merisiers Saint-Aubin, BP 48, F-91192 Gif-sur-Yvette, France
3ISMN-CNR, via Gobetti 101, Bologna, Italy
4Laboratorio Nazionale TASC, INFM-CNR, in Area Science Park, S.S. 14, Km 163.5, I-34012 Trieste, Italy
5Dipartimento di Fisica, Università di Modena e Reggio Emilia, Via A. Campi 231/A, I-41100 Modena, Italy
6European Synchrotron Radiation Facility, BP 220, F-38043 Grenoble, France
共Received 18 November 2009; revised manuscript received 22 December 2009; published 27 January 2010
兲
We demonstrate that the growth of Fe/共Ga,Mn兲As heterointerfaces can be efficiently controlled by epitaxy and that robust ferromagnetism of the interfacial Mn atoms is induced at room temperature by the proximity effect. X-ray magnetic circular dichroism and x-ray resonant reflectivity data, supported by theoretical calcu- lations, were used to monitor both temperature and magnetic field dependence of the Mn magnetic moment in the semiconducting host. We identify distinct Mn populations, each of them with specific magnetic character.DOI:10.1103/PhysRevB.81.035211 PACS number共s兲: 75.50.Pp, 71.20.Nr, 78.70.Dm
I. INTRODUCTION
Diluted magnetic semiconductors
共DMS兲
hold promises for integrating spin control in electronic devices.1,2Although the correlation between magnetic and transport properties in DMS is known to be a crucial ingredient toward possible applications, the physical mechanisms governing both the magnetic character and the magnetization alignment of the metallic dopants in the semiconducting environment are open challenges of present DMS research.3–5One further key subject arises from the Curie temperature, presently around 200 K in the most representative DMS material共Ga,Mn兲As 共Ref.
6兲 while ferromagnetic共FM兲
behavior well beyond room temperature共RT兲
would be required in future spintron- ics devices. A promising direction that goes beyond conven- tional methodologies is to tailor novel properties by exploit- ing interface effects in highly controlled heterostructures共HS兲,
as extensively demonstrated in oxide-based materials.7–9 Following this approach, recent experimental efforts aimed at tuning specific magnetic properties in FM metal/DMS-based interfaces. In the case of direct exchange- coupled HS, the investigation of spin valve effect in MnAs/共Ga,Mn兲As at low temperature 共4.2 K兲
found the apparent formation of a inhomogeneous magnetic spring in the共
Ga,Mn兲
As region10 and a decrease in exchange coupling after insertion of a spacer layer关MnAs/
p-GaAs/共Ga, Mn兲As兴.11Differently, HS with 3dFM metals revealed共i兲
independent magnetization switching in共Ni
80Fe20兲
/共Ga, Mn兲As, despite direct contact
12 and共ii兲
FM behavior of Mn at RT at nonepitaxial Fe/共
Ga,Mn兲
As inter- faces, with antiparallel alignment of the Fe and Mn moments.13 A firm understanding of the mechanisms in- volved in these effects requires on one side a full control of HS growth and on the other side the ability to probe the electronic and magnetic properties in a chemical and depth sensitive way. Here we report polarization-dependent x-ray experiments on fully epitaxial Fe/共Ga,Mn兲As HS, where we are able to follow the field, temperature, and depth evolution of the different magnetic configurations of the Mn ions. Ourresults reveal the presence of a electronic configuration of the Mn ions under the proximity effect of the Fe overlayers, indicating that
共a兲
Mn hybridization is playing a fundamental role in the magnetic properties of the system and共b兲
para- magnetic and ferromagnetic Mn共
antiparallel to Fe兲
have dis- tinct spectroscopic fingerprints in an applied magnetic field, i.e., the different magnetic behavior in a sizeable part of the Mn atoms at the interface共ⲏ1 nm兲
is “switched on” by the Fe overlayer.II. EXPERIMENTAL DETAILS
The ferromagnetic
共Ga,Mn兲As samples 共Mn 3 – 8 % dop-
ing, 50–150 nm thick兲have been grown by molecular-beam epitaxy共MBE兲. The rate for the low-temperature growth was
about 0.6 Å/s共
230° – 250 ° C兲
. Subsequently the samples were transferred to a metal MBE for Fe evaporation. An optimized protocol to obtain contamination free and ordered surfaces was used, by combining low-temperature ultrahigh vacuum annealing共150 ° C兲
and soft Ar+sputtering共500 eV,
45° incidence angle兲
for a typical time of 30 min. Details are given elsewhere.14,15 Reference samples were obtained by HCl etching. The Fe growth rate was controlled in situ by quartz monitors and the Fe overlayers were prepared as stepped wedges with discrete thickness values to ensure that the growth conditions were identical for all thicknesses in the covered range from 0 to 23 ML. The resulting well ordered surfaces were verified by reflective high-energy electron dif- fraction. Samples were covered by a 8-nm-thick epitaxial Au共
001兲
layer to prevent oxidation. The magnetic anisotropy axis of both Fe and共Ga,Mn兲As layer, determined by a super-
conducting quantum interference device magnetometer and by magneto-optical Kerr effect共MOKE兲, closely resembles
Fe/GaAs共001兲and the nontreated共Ga,Mn兲As/GaAs共001兲, re-
spectively, excluding sizeable variations in the magnetic properties after the Ar+ bombardment.14–16 Moreover, the Curie temperature of the sputtered共Ga,Mn兲As is identical to
the one of the as-grown samples共T
c= 67⫾3 K兲. Figure1共a兲 presents MOKE results. At the wavelength used the Kerrcontrast for
共Ga,Mn兲As and Fe has opposite sign which
makes it easy to separate the switching behavior of the two different layers. We show data at 300 K where only the Fe film produces a magnetic contrast and at 6.5 K for the Fe/GaMnAs film and for a pure
共Ga,Mn兲As film for compari-
son. We acquired x-ray absorption spectroscopy共XAS兲
and x-ray circular magnetic dichroism共XMCD兲
spectra at the Mn and Fe L edges at the ID08 dragon beamline共ESRF,
France兲
and at the APE beamline of the Elettra Synchrotron共Trieste, Italy兲
in a temperature range of 10–300 K and a base vacuum ⬍5⫻10−10 mbar. Total electron yield共TEY兲
and photon yield共PY, bulk sensitivity
⬎40 nm兲 were ac- quired simultaneously. All spectra were normalized by theintensity of the incident beam. The TEY Fe XAS signal has been corrected for saturation effects, whereas PY mode has been preferred for Mn, due to both the need of bulk sensi- tivity and to the dilution in GaAs, resulting in negligible self-absorption effects of the fluorescence signal. X-ray reso- nant magnetic scattering
共XRMS兲
measurements have been carried out at ID08 in specular reflection geometry by fixing the incoming photon energy at the L3Mn absorption peak and scanning the grazing angle from 0° to 40°. Right cir- cularly polarized radiation was used and reflectivity mea- surements were performed in remanence after having applied the magnetic field in opposite directions. Simulation and fit- ting of the x-ray resonant reflectivity measurements havenormalizedKerrintensity(arb.units)
-1000 0 1000
H (Oe)
(a)
Fe/(Ga,Mn)As T=6.5 K
Fe/(Ga,Mn)As T=300 K
(Ga,Mn)As T=6.5 K
Hc,Fe=110 Oe
Hc,(Ga,Mn)As=55 Oe
Hc,Fe=207 Oe Hc,(Ga,Mn)As=80 Oe
20 10 0
TEY-XMCD(%)
740 730 720 710 700
Photon energy (eV) 0.6
0.4
0.2
0.0
TEY-XAS(arb.units)
(c) P+, M+
P+, M-
Fe L2,3
(b)
(Ga,Mn)As Fe Au
2 nm
0.4
0.2
0.0
XAS(arb.units)
655 650 645 640
Photon energy (eV)
0 -1 -2
PY-XMCD(%)
dE=0.5eV
(d)
PY TEY
Fet= 23ML Fet= 0ML
Mn L2,3
FIG. 1. 共Color online兲 共a兲Normalized MOKE data measured at 300 and 6.5 K for Fe/共Ga,Mn兲As and 共Ga,Mn兲As. The Kerr contrast 共measured with= 632 nm兲is opposite for Fe and共Ga,Mn兲As which makes it possible to easily distinguish the two layers. The coercive field of Fe increases when going to lower temperatures as expected. A slight increase inHcis observed for the pure共Ga,Mn兲As when Fe is deposited on top.共Panel b兲High-resolution cross-sectional TEM image of the epitaxial Fe共2 nm兲/共Ga,Mn兲As共50 nm兲sample, capped with 8 nm of Au. The interface region is sharp and no intermixing is visible. The black horizontal line共bottom right兲 indicates the scale共2 nm兲. 共Panel c兲 FeL2,3 edge spectra as measured in TEY mode 共T= 120 K , H= 0.04 T , Fe= 23 ML兲, with circular polarization 共P+兲 and opposite magnetization共M+,M−兲. Black arrows indicate the XAS共left兲and XMCD共right兲scale, respectively.共Panel d兲MnL2,3edge spectra as measured in TEY and PY mode共T= 120 K , H= 0.04 T , Fe= 23 ML兲. Black arrows indicate the XAS共left兲and XMCD共right兲scale.
Mn XAS spectra acquired in PY mode 共blue/dark gray continuous line兲 and TEY mode 共red/gray points兲 differ by 0.5 eV and display different fine structure atL3edge. XMCD signal at the MnL3edge共green/light gray curve兲is observed, indicating antiparallel magnetic coupling of Mn with respect to the Fe. No XMCD signal is measured in absence of the Fe overlayer共yellow/light gray curve in panel d, Fe= 0 ML兲.
been performed using the Pythonic Programming for Multilayer
共
PPM兲
software.17,18III. RESULTS AND DISCUSSION
Figure 1 depicts the magnetic and structural characteris- tics of the Fe/共Ga,Mn兲As HS. The crystallinity of the Au/Fe/
共
GaMn兲
As stack is confirmed by the transmission electron microscopy共TEM兲
image共panel b兲, revealing a sharp inter-
face between Fe and共Ga,Mn兲As. Reference XAS-XMCD
signals from the FeL2,3and MnL2,3edges are displayed in panels共
c兲
and共
d兲
, for 23 ML Fe and 0.04 T at 120 K, i.e., well aboveTcof the共Ga,Mn兲As substrate. The PY spectrum
is located at⬵0.5 eV lower binding energy 共BE兲
and it is less structured than the TEY one. The PY-XMCD displays FM behavior and antiparallel共AP兲
Mn orientation.13 We stress that, above Tc, a XMCD signal at MnL2,3 edge is measured only in presence of the Fe overlayer. The yellow curve in panel共d兲
shows no XMCD at the Mn edge in the region free of Fe, thus indicating a magnetic behavior switched on by the proximity effect of the Fe film.From spectroscopic results, it is generally agreed that the two prevailing Mn electronic structures in
共Ga,Mn兲As are
MnGa共Mn substitutional on the Ga site兲
and Mn forming oxides or aggregates near the surface, with a hybridd4-d5-d6 and a purelyd5 electronic configuration共
localized Mn兲
, re- spectively. The latter configuration has a more structured XAS spectrum, located at higher BE with respect to the d4-d5-d6 one.19–22A recent detailed study on the depth con- centration of Mn in共Ga,Mn兲As reveals the presence of Mn
withd5configuration up to 6 nm from the interface, coexist- ing with MnGa and not related to oxide.23 Results in Fig.1, obtained on controlled epitaxial systems with negligible ox- ide contribution indicate a surface-interface region with d5-like character共TEY spectrum兲
and a bulk region, mainly representative of the MnGa共
PY spectrum兲
. For the sake of simplicity, we will refer in the following to Mn-1共Mn
Gasubstitutional兲, Mn-2
共d
5-like Mn兲, and to Mn-AP共FM Mn
antiparallel to Fe兲. The evolution of the MnL2,3XMCD sig- nal vs applied magnetic field is presented in Fig. 2共a兲共
T= 120 K, Fe 23 ML兲
. At low field共
0.04 T兲
only the Mn-AP dichroic component is visible at the L3 edge, con- firming previous results.13 Increasing the field, further com- ponents appear at lower BE, evolving in a structured line shape with a double peak at the L2 edge and with the mag- netization aligned parallel to the field. These components correspond to the paramagnetic contribution of Mn-1. The XAS/XMCD arising from Mn-1 has been associated to a mixed-valence ground state due to the large Mn 3d hybrid- ization with the surrounding 4spstates.19,24,25Numerical cal- culations indicated a mixed 80% d5-20% d6L ground state共L
is a ligand hole兲.24,25 In Mn-AP, both the energy shift of the XMCD minimum and the line shape suggest a different valence state with lower occupancy of the 3d levels, imply- ing a lowering of the d6 character and increment of thed5 one.In Fig.2共b兲the Mn-AP XMCD at 0.04 T is compared to atomic multiplet calculations, namely, a pure Mnd5ground- state configuration based on the ligand field model, for an
isolated ion in spherical symmetry,24,26,27 and a d5
共50%兲-d
6共50%兲
mixed-valence configuration. The d5共50%兲-d
6共50%兲
mixed-valence XMCD has been calculated within the impurity Anderson model with the ad- ditional parameters: Eg共
d6兲
= −3 eV, Ef共
d6兲
= −4 eV, and Tt2g= 2Teg= 2. Eg and Ef are related to the charge-transfer energies for ground and excited states whileTt2gandTegare the charge-transfer integrals for t2g and eg orbitals, respectively.28 Introducing in the calculation furtherd4char- acter ord4-d5 mixing withoutd6terms does not provide any better agreement to the experimental Mn XMCD. The com- parison to the calculation highlights the smearing of the mul- tiplet features in the experimental results and reveals unam- biguously that Mn-AP XMCD does not bear a pure atomiclike character, suggesting a mixed-valence state with itinerant character. Moreover, the broadened features at the high-energy side of theL3XMCD that appear for a metalli- clike condition are totally absent, thus excluding Mn alloying/clustering. The observed mixed-valence features suggest a rearrangement of the local density of states associ- ated to changes in the local Mn environment that may be interpreted by the formation of an impurity band, similar to half-metallic systems, where Mn atoms have localized mag- netic moments but delocalized 3d electrons.29,30To disentangle the Mn-AP ferromagnetic character from the paramagnetic one, we now need to assign the different contributions to the XMCD spectra. To this aim, a fitting procedure was performed, using a linear combination of ex-
XMCD-PY(arb.units)
655 650 645 640
Photon Energy (eV) (x4)
(a)
Mn-AP Mn-1
5 T 3 T 2 T 1.5 T 0.5 T 0.04T
XMCD(arb.units)
655 650 645 640
Photon Energy (eV) (b) Calc. d5-d6L
Calc. d5
Exp.
Mn L2,3XMCD
FIG. 2. 共Color online兲 共Panel a兲 Evolution of the Mn XMCD line shape vs applied magnetic field 共T= 120 K , Fe= 23 ML兲, measured in PY mode. The inset shows a sketch of the experimental geometry. 共Panel b兲 Comparison of calculated and experimental 共120 K, 0.04 T兲Mn-AP XMCD. All curves have been normalized to the minimum of the experimental XMCD and are arbitrarily shifted in energy to align the theoretical to the experimental result. Features related to multiplet splitting effects are absent or smeared, as indi- cated by the vertical dashed lines. The interatomic screening and excessive electron-electron repulsion of the free ion were taken into account by reducing the Slater integrals to 80%. The 2pspin-orbit interaction was scaled to 103%. An exchange field of 0.01 eV was applied along the magnetization direction. The calculated spectra include a Lorentzian of 0.25共0.4兲eV for theL3共L2兲edge to account for intrinsic linewidth broadening and a Gaussian of= 0.1 eV for instrumental broadening. For sake of comparison, all calculated spectra are normalized to the same amplitude and arbitrarily aligned in binding energy.
perimental Mn-1,2 XAS and XMCD spectra from Ref. 22, constrained to reproduce XAS and XMCD data vs magnetic field, temperature, and Fe thickness. The Mn-AP XMCD cannot be reproduced by the superposition of Mn-1 and Mn-2, and has to be explicitly included in the fitting routine.
Results for 120 K, 1 T, and 23 ML of Fe are presented in Fig.
3共a兲. We find only two contributions in the XMCD spectra:
paramagnetic from bulk Mn-1
共T
⬎Tc兲
and ferromagnetic from Mn-AP. Although a 1 T field is applied, no paramag- netic contribution associated to Mn-2 character is found, thus confirming the absence of Mn oxide in the epitaxial HS.Disentangling the different Mn character in XMCD spectra allows tracing the evolution vs magnetic field, i.e., to recon- struct site selective
共Mn-1 and Mn-AP兲
hysteresis loops, aspresented in Fig.3
共
b兲
. The Mn-AP hysteresis matches the Fe one up to 2 T, indicating the robustness of the magnetic coupling. At higher magnetic fields the observed partial re- duction in the Mn magnetic moment共⬵40%兲
suggests a re- orientation of the Mn magnetization in a direction parallel to the field. Having identified the novel magnetic and electronic behavior of Mn-AP, we now address the distribution profile of Mn-AP across the interface, by means of XRMS results.Figure 4
共
a兲
reports the-2 scan obtained at h= 640.4 eV andT= 138 K, averaging the reflectivity curves for parallel and antiparallel relative orientation of photon helicity and magnetic field. Such a signal is insensitive to the magnetic moments of the sample and depends only upon the electron charge distribution and chemical properties. In Fig. 4共
b兲
we report the asymmetry ratio AR=共r↑ ↑
−r↑↓兲/共r
↑ ↑−r↑↓兲, where共r↑↑兲
and共r↑↓兲
are the reflectivity values measured for parallel and antiparallel orientation of the photon helicity and magnetization, respectively. The AR value gives infor- mation about the distribution of the Mn magnetic moment in the sample. We fitted the-2 scan using the thickness and roughness of the various layers关Au,Fe,共GaMn兲As,GaAs兴
as fitting parameters. Absorption coefficients for the Mn species have been obtained experimentally from XAS measure- ments. The real part of the refractive index is obtained from the imaginary part, which is directly proportional to the ab- sorption coefficient, through Kramers-Kroenig transforma- tion. The fitting results well reproduce the experimental curve共
the values of the best fit parameters are listed in the caption of Fig. 4兲, in good agreement with TEM results.Considering a distribution of Mn atoms antiparallely oriented with respect to Fe, we are able to reproduce maxima and minima of the measured oscillations: best agreement is found
1.2 0.8 0.4 XAS-PY(arb.units) 0.0
655 650 645 640
Photon energy (eV)
-5 0 5 10
XMCD(%)
(a)
Mn-1 Mn-2 Mn-AP Experiment Fitting result
20
10
0
-10
-20
Fe,Mn-1XMCD(%)
-4 -2 0 2 4
Magnetic Field (Tesla) -2 -1 0 1 2
Mn-APXMCD(%)
(b)
Mn-1 Mn-AP Fe (x0.5)
FIG. 3. 共Color online兲 共a兲 Fitting results and experimental MnL2,3 XAS/XMCD spectra 共black dashed curve and red/gray curve, respectively兲 using template spectra from Ref. 21 共Mn-1, blue/dark gray dashed curve and Mn-2 yellow/light gray dashed curve兲 and experimental Mn-AP XMCD 共green/light gray dot- dashed curve兲. Experimental curves correspond tot共Fe兲= 23 ML, T= 120 K, and H= 1 T. Best agreement between experimental XMCD and fit is obtained by using only Mn-1 and Mn-AP compo- nent.共b兲Hystereses cycles traced from the XMCDL3peak height at the Mn and Fe edges, in percentage of the XAS sum peak 共I+−I−兲/2, forT= 120 K, Fe= 23 ML. Black arrows indicate the Fe XMCD scale 共left兲 and Mn XMCD one 共right兲. The Mn-1 共blue/
dark gray triangles兲and Mn-AP共green/light gray circles兲hystereses were extracted by fitting the L3 XMCD spectra at each magnetic field, using the same fitting procedure of panel共a兲 共linear combina- tion of Mn-1 and Mn-AP components兲. The resulting Mn-1 behav- ior is purely paramagnetic, as expected above Tc of 共Ga,Mn兲As.
Experimental values of Fe XMCD 共red/gray diamonds兲 are multi- plied by 0.5 for sake of comparison.
100 101 102 103 104
Reflectivity(Arb.Units)
40 35 30 25 20 15 10 5
Angle (degrees) 15x10-3
10 5 0 -5 -10
NormalizedAsymmetry
(a)
(b)
Fit:
Exp:
Fit: A B
Exp:
Magnetization(Arb.Units)
-25 0 25
z (Å)
Au Fe
(c)
(d)
Ga1-xMnxAs Mn-APmagnetization profiles
A B
FIG. 4. 共Color online兲 共a兲 Specular x-ray reflectivity at 138 K collected with a photon energy of 640.4 eV 共black circles兲 com- pared with fitting results obtained with the PPM program共red/gray line兲. Values used in the fitting are共thickness and roughness in Å, respectively兲 共GaMn兲As, 369 and 3.0. Fe 27 and 4.1. Au 41 and 2.1.
GaAs substrate with roughness 1.0 Å.共b兲Experimental asymmetry ratio AR共blue/dark gray circles兲, as defined in the text. Comparison of the experimental AR with two simulated magnetic profiles, i.e., a 10-Å-thick magnetic layer of Mn ions located in the 共Ga,Mn兲As environment共red/gray curve, panel d兲or in the Fe film共green/light gray curve, panel d兲.共c兲Sketch of the experimental geometry used for XRMS experiments.
using a 10-Å-thick layer of Mn atoms with a constant step- like magnetic profile in
共Ga,Mn兲As. We also simulated the
behavior of the same Mn thickness, but with Mn diffused into the Fe layer, in antiparallel configuration. The period of oscillations is almost equal but the positions of the maxima and minima are inverted. This excludes a sizeable presence of magnetic Mn atoms into the Fe film. We stress that the simple steplike magnetization profile used in the simulation, although able to reproduce the XRMS data, should be con- sidered as a lower estimate of the Mn-AP magnetic depth profile.IV. CONCLUSIONS
In conclusion, by the use of a broad range of chemical sensitive spectroscopic tools we have provided a complete description of the Mn magnetic properties in fully epitaxial Fe/GaMnAs HS. The use of the magnetic proximity effect between Fe and Mn made it possible to reveal a novel elec- tronic configuration of the interfacial Mn atoms, with differ- ent magnetic behavior with respect to the bulk ones. Our
original approach demonstrates that important insights in the fundamental interactions driving ferromagnetism in DMS- based systems can only be obtained by tailoring novel prop- erties in controlled heterostructures. Future efforts in the di- rection of hybrid metal/DMS ferromagnetic structures may achieve the control of switching fields and Curie tempera- ture, opening a completely new avenue for the design of hybrid metal/DMS/semiconductor structures for future spin- tronic devices.
ACKNOWLEDGMENTS
Financial support by the DFG through the SFB 689 is gratefully acknowledged. This work has been partially funded by CNR-INFM. Thanks are due to Alessandro Mirone for fruitful discussion and support for the use of the
PPMcode, and to Stefano Nannarone and Bruce A. Davidson for fruitful discussions on XRMS technique. A.V. was sup- ported by the FVG Regional Project SPINOX funded by Legge Regionale 26/2005 and the decreto 2007/LAVFOR/
1461.
*panaccioneg@elettra.trieste.it
1S. A. Wolf, D. D. Awschalom, R. A. Buhrman, J. M. Daughton, S. von Molnár, M. L. Roukes, A. Y. Chtchelkanova, and D. M.
Treger, Science 294, 1488共2001兲.
2D. D. Awschalom and M. E. Flatté, Nat. Phys. 3, 153共2007兲.
3A. H. Macdonald, P. Schiffer, and N. Samarth, Nature Mater. 4, 195共2005兲.
4I. Zutic, J. Fabian, and S. Das Sarma, Rev. Mod. Phys. 76, 323 共2004兲.
5G. Wastlbauer and J. A. C. Bland, Adv. Phys. 54, 137共2005兲, and references therein.
6K. Olejnik, M. H. S. Owen, V. Novák, J. Mašek, A. C. Irvine, J.
Wunderlich, and T. Jungwirth, Phys. Rev. B 78, 054403共2008兲; J. Cryst. Growth 311, 2151共2009兲.
7J. Chakhalian, W. Freeland, H.-U. Habermeier, G. Cristiani, G.
Khaliullin, M. van Veenendaal, and B. Keimer, Science 318, 1114共2007兲.
8H. Yamada, Y. Ogawa, Y. Ishii, H. Sato, M. Kawasaki, H. Akoh, and Y. Tokura, Science 305, 646共2004兲.
9C. Chappert, A. Fert, and F. Nguyen Van Dau, Nature Mater. 6, 813共2007兲.
10M. Zhu, M. J. Wilson, B. L. Sheu, P. Mitra, P. Schiffer, and N.
Samarth, Appl. Phys. Lett. 91, 192503共2007兲.
11M. Zhu, M. J. Wilson, P. Mitra, P. Schiffer, and N. Samarth, Phys. Rev. B 78, 195307共2008兲.
12S. Mark, C. Gould, K. Pappert, J. Wenisch, K. Brunner, G.
Schmidt, and L. W. Molenkamp, Phys. Rev. Lett. 103, 017204 共2009兲.
13F. Maccherozzi, M. Sperl, G. Panaccione, J. Minár, S. Polesya, H. Ebert, U. Wurstbauer, M. Hochstrasser, G. Rossi, G. Wolters- dorf, W. Wegscheider, and C. H. Back, Phys. Rev. Lett. 101, 267201共2008兲.
14F. Maccherozzi, G. Panaccione, G. Rossi, M. Hochstrasser, M.
Sperl, M. Reinwald, G. Woltersdorf, W. Wegscheider, and C. H.
Back, Surf. Sci. 601, 4283共2007兲.
15F. Maccherozzi, G. Panaccione, G. Rossi, M. Hochstrasser, M.
Sperl, M. Reinwald, G. Woltersdorf, W. Wegscheider, and C. H.
Back, Phys. Rev. B 74, 104421共2006兲.
16Fe growth on etched substrates results in a polycrystalline mor- phology, with high degree of magnetic disorder. In this case, no detectable XMCD at the Mn edge has been found.
17L. G. Parratt, Phys. Rev. 95, 359共1954兲PPMcode is available at http://www.esrf.eu/computing/scientific/PPM/ppm.html
18B. L. Henke, E. M. Gullikson, and J. C. Davis, At. Data Nucl.
Data Tables 54, 181共1993兲.
19K. W. Edmonds, P. Bogusławski, K. Y. Wang, R. P. Campion, S.
N. Novikov, N. R. S. Farley, B. L. Gallagher, C. T. Foxon, M.
Sawicki, T. Dietl, M. Buongiorno Nardelli, and J. Bernholc, Phys. Rev. Lett. 92, 037201共2004兲.
20K. W. Edmonds, W. Edmonds, N. R. S. Farley, R. P. Campion, C. T. Foxon, B. L. Gallagher, T. K. Johal, G. van der Laan, M.
MacKenzie, J. N. Chapman, and E. Arenholz, Appl. Phys. Lett.
84, 4065共2004兲.
21A. A. Freeman, K. W. Edmonds, N. R. S. Farley, S. V. Novikov, R. P. Campion, C. T. Foxon, B. L. Gallagher, E. Sarigiannidou, and G. van der Laan, Phys. Rev. B 76, 081201共R兲 共2007兲.
22K. W. Edmonds, G. van der Laan, A. Freeman, N. R. S. Farley, T. K. Johal, R. P. Campion, C. T. Foxon, B. L. Gallagher, and E.
Arenholz, Phys. Rev. Lett. 96, 117207共2006兲.
23F. Kronast, R. Ovsyannikov, A. Vollmer, H. A. Dürr, W. Eber- hardt, P. Imperia, D. Schmitz, G. M. Schott, C. Ruester, C.
Gould, G. Schmidt, K. Brunner, M. Sawicki, and L. W. Molen- kamp, Phys. Rev. B 74, 235213共2006兲.
24G. van der Laan and B. T. Thole, Phys. Rev. B 43, 13401共1991兲.
25H. Ohldag, V. Solinus, F. U. Hillebrecht, J. B. Goedkoop, M.
Finazzi, F. Matsukura, and H. Ohno, Appl. Phys. Lett. 76, 2928 共2000兲.
26T. Cowan,The Theory of Atomic Structure and Spectra共Univer-
sity of California Press, Berkeley, CA, 1981兲.
27Systematic investigation of crystal-field effects that could lead to such XMCD curve did not produce better agreement with the experimental XMCD. Therefore the theoretical XMCD for a pure Mnd5ground state in Fig.2was obtained in O共3兲spherical symmetry.
28F. M. F. de Groot, Coord. Chem. Rev. 249, 31共2005兲.
29J. Kübler, A. R. Williams, and C. B. Sommers, Phys. Rev. B 28, 1745共1983兲.
30N. D. Telling, P. S. Keatley, G. van der Laan, R. J. Hicken, E. Arenholz, Y. Sakuraba, M. Oogane, Y. Ando, K. Takanashi, A. Sakuma, and T. Miyazaki, Phys. Rev. B 78, 184438 共2008兲.